Supporting decision-making with organizational decision memory.

Abstract

Although organizational memory remains a relatively ill-defined construct, information systems developers seem anxious to build software that is intended to facilitate its development and use. The construct seems to consist of concepts from organizational theory, information systems design, and decision support. Decision support is one of the commonly proposed uses of organizational memory, and a common justification for efforts to build information systems to support organizational memory. Yet, the broad range of information that might be stored in and recalled from an organization's memory makes it difficult to define requirements for an automated organizational memory system. This dissertation describes the design and implementation of a laboratory experiment aimed at discovering whether information about historic decision making behavior could be useful to decision makers in organizations. Several hypotheses were tested. The first is that information about decision-making behavior from the past will lead decision makers to make better decisions in the present. Second, the experiment tested the hypothesis that historic information from an organizational (collective) source would be more helpful to decision makers than similar information from an individual source. These hypotheses were supported. Third, it was hypothesized that historic decision information, presented in the format of a linear regression model would be more helpful to decision makers than similar information in a textual format. This hypothesis was not supported. Additionally, hypotheses were tested concerning the relative benefits of collective and individual sources of historic information versus no historic information, and comparing the benefits of the regression format and textual format versus no historic information. The findings indicated that the information from a collective source provided better decision support than no historic information, but information from the individual source was not significantly superior to no historic information. Further, the information in the textual format was significantly better than no information, but the regression format was not significantly better than no historic information. In general, the findings indicate that historic decision information can lead decision makers to make better decisions, but source of information and the presentation format are important variables affecting the extent of the benefit.